Collaborative Stacked Denoising Auto-Encoders for Refining Student Performance Data

摘要

The Cognitive modelling can discover the latent skills of students for predicting their performance on each problem and formulate personalized remedy recommendation. However, Uncovering precise student performance data without noise is a difficult task. In order to solve this problem, this paper proposes a method based on auto-encoder to refine student response data, which can obtain response data without noise. We combine educational hypotheses to the model by adding Q matrix constraint in it. Further, The experimental results show that our proposed method has better performance in refining the original student response data.